ASSESSING THE EFFECTS OF LAND USE CHANGE ON RUNOFF IN BEDOG SUB WATERSHED YOGYAKARTA | Prasena | Indonesian Journal of Geography 2406 4066 1 SM
ISSN 0024-9521
IJG Vol. 45, No.1, June 2013 (48 - 61)
© 2013 Faculty of Geography UGM and
The Indonesian Geographers Association
ASSESSING THE EFFECTS OF LAND USE CHANGE ON RUNOFF
IN BEDOG SUB WATERSHED YOGYAKARTA
Aris Prasena
prasena.aris@gmail.com
BAPPRDA PROV DIY kompleks Kepatihan Danurejan, Yogyakarta
D.B. Pikha Shrestha
shrestha@itc.nl
Faculty of Geo-Information and Earth Observation, University Of Twente
ABSTRACT
The study was to assess the effects of land use change on runoff in the Bedog sub watershed.
Soil and Water Assessment Tool-Water Balance (SWAT-WB) hydrological modeling was used
to predict runoff for years of 2001, 2006, and 2010. Land use in Bedog has rapidly changed
in last few decades due to agglomeration process in Yogyakarta City. Coverage of mixed
garden decreased during period of 2001-2010 despite still as predominant land use in total.
On the other side, change detection analysis revealed that there was an increase of
settlements coverage from 9.51% to 13.79% in the same period. Sensitivity analysis revealed
that soil properties were the most sensitive parameters on runoff generation. Calibration was
performed for years of 2001, 2006, and 2010 and the result shows an acceptable
performance in runoff simulation. Changes in land use were responsible for an increase in
the annual runoff between 3.42% 4.67%. This study showed that dynamics of runoff can be
predicted by forecasting and simulating future land use.
Key Words: Remote sensing, Geographic Information System (GIS), land use change, Soil and
Water Assessment Tool-Water Balance (SWAT-WB), runoff.
ABSTRAK
Penelitian ini bertujuan untuk menganalisis dampak dari perubahan penggunaan lahan
terhadap limpasan di sub DAS Bedog. Pemodelan hidrologi berupa Soil and Water
Assessment Tool-Water Balance (SWAT-WB) digunakan untuk memprediksi limpasan
selama tahun 2001, 2006, dan 2010. Penggunaan lahan di Bedog dengan cepat berubah
dalam beberapa dekade terakhir akibat proses aglomerasi di Kota Yogyakarta. Lahan
berupa kebun campuran berkurang selama periode 2001-2010, meskipun lahan tersebut
masih merupakan penggunaan lahan dominan secara keseluruhan. Di sisi lain, analisis
deteksi perubahan menunjukan bahwa terjadi peningkatan cakupan permukiman dari 9,51%
menjadi 13,79% pada periode yang sama. Analisis sensitivitas menunjukkan bahwa sifat
tanah merupakan parameter yang paling sensitif dalam menghasilkan limpasan. Kalibrasi
yang dilakukan selama bertahun-tahun 2001, 2006, dan 2010 menunjukkan kinerja yang
dapat diterima dalam simulasi limpasan. Perubahan penggunaan lahan berdampak pada
peningkatan limpasan tahunan antara 3,42% - 4,67%. Studi ini menunjukkan bahwa
dinamika limpasan dapat diprediksi dengan peramalan dan simulasi penggunaan lahan di
masa depan.
Kata Kunci: Penginderaan jauh, Sistem Informasi Geografis (GIS), perubahan penggunaan
lahan, Soil and Water Assessment Tool-Water Balance (SWAT-WB), limpasan.
Indonesian Journal of Geography, Vol 45, No.1, June 2013 : 48 - 61
INTRODUCTION
Changes in land use and land cover have
occurred from prehistoric times till present
[Ellis, 2010]. Mankind has changed land
surface in attempts to improve the
availability of space, and enhance security
of basic natural resources. The phenomena
of land use changes and its consequences
to the environment are important to be
observed in attempts to implementing
sustainable development.
Agglomeration of Yogyakarta as the
centre of economic activities and
education services in Yogyakarta Province
has witnessed rapid development of
buildings, infrastructure, and changes of
land use from non-urban to urban features.
As a consequence land use change took
place, growth of settlement in Bantul,
Sleman, and Yogyakarta increased 7.16%
in period of 1994-2000 while on the other
hand areas of paddy field decreased from
46.31 % to 39.49 % [BAPPEDA of
Yogyakarta Province, 2007].
Figure 1. Map of Progo watershed
As consequence, extensive changes from
non-urban into urban features possibly
generate higher runoff that will increase
flood risk both in the area of Yogyakarta
agglomeration and in the lower part of
watershed.
Bedog is part of Progo watershed which
one of urbanized watersheds (figure 1).
The middle stream of Bedog is part of
agglomeration
Yogyakarta
which
determined as national growth pole. It has
role as centre of economic and education
activities. On the other hand, most part of
upper area and downstream has different
characteristic as an agriculture areas that
contribute in food security program.
The main objective of this research is to
assess the effects of land use change on
runoff in Bedog sub watershed,
Yogyakarta Province, Indonesia.
The specific objectives of this research
are:
1. To identify the existing land use
pattern.
2. To analyze pattern and magnitude
of land use change 2001-2010.
3. To determine runoff using a model.
4. To analyze the relation between
land use change and runoff
dynamic.
Middle part of Bedog as urban fringe of
Yogyakarta has witnessed urbanization
and changes in land use.
Entering the 20th century, land use change
has become the world’s problem closely
related with other global issues such as
population growth, food security, and
climate change. It is estimated that builtup or impervious areas occupy between
2% to 3% of the Earth’s land surface
49
ASSESSING THE EFFECTS
Aris Prasena, D.B. Pikha Shrestha
vegetation, soil, lakes, rivers, snowfields
and oceans.
[Lambin and Geist, 2006]. This is also
supported by deforestation in tropical
countries [Pagiola, 2000] and extensively
application of agricultural intensification
in productive land using fertilizers,
pesticides and irrigation systems in recent
decades.
Precipitation which covers all forms of
water being released by the atmosphere is
the beginning of a whole chain of events
that occurs in watersheds and the source of
water that replenishes soil moisture,
stream flows, lakes, glaciers, etc [Salas,
2006]. According to Washington State
Department
of
Ecology,
[1986]
precipitation falls on the earth and either
percolates into the soil or flows across the
ground.
In Indonesia, the loss of natural forest in
Sumatra, Kalimantan, Sulawesi, and Irian
Jaya between 1985 and 1997 is over 19
million ha of forest, including 6.7 million
ha in Sumatra and 8.5 million ha in
Kalimantan [Pagiola, 2000]. This amount
is an average annual rate in those two
islands of 1.26 million ha per year.
The term runoff incorporates the
movement of liquid water above and
below the surface of the earth [Davie,
2008]. Surface runoff is part of the rainfall
that flows over the surface of the soil to
the rivers, lakes, and oceans [Asdak,
1995]. Runoff mechanisms that contribute
to stream flow can be distinguished as
overland flow, through flow/lateral flow,
and groundwater flow [Davie, 2008].
Overland flow is the water which runs
across the surface of the land before
reaching the stream. Lateral flow is the
water which runs the subsurface, occurs in
the shallow subsurface, predominantly,
although not always, in the unsaturated
zone. Groundwater flow is in the deeper
saturated zone.
Yunus,
[2008]
points
out
that
industrialization has become main driving
force for rapid urbanization since most of
manufacturing industries tend to be
located around city centre to minimize
production cost. Lambin and Geist, [2006]
argue that urbanization affects land in rural
areas through the ecological footprint of
cities includes, agricultural land in periurban areas for residential, infrastructure,
and amenity uses, which blurs the
distinction between cities and rural.
Land use change due to urban growth has
occurred in Yogyakarta Province. Konig et
al., [2010] point out that population
density in the Yogyakarta increased by
84% (from 532 to 979 persons per km2) in
1970-2000. This rapid growth resulted in
urban-rural expansions of built-up areas by
13% (1990–2006). In period of 19932006, growth of built-up areas and new
rural settlements was doubled, while
agricultural land decreased by 25%.
Factors affecting runoff are factors related
to climate, particularly rainfall and
characteristics of watershed [Asdak, 1995].
The rate and volume of runoff is
determined by duration, intensity, and
distribution of rainfall while characteristics
of watershed are shape, morphometry,
topography, geology, and land use (type
and vegetation density).
Hydrological cycle is a conceptual model
of how water moves around between the
earth and atmosphere in different states as
a gas, liquid or solid [Davie, 2008].
Washington State Department of Ecology
[1986] suggests that the hydrologic cycle
begins when water evaporates into the
atmosphere that can be derived from
Land use change plays role as main factor
in hydrological cycle. Giertz et al., [2004]
point out that there are effects of land use
change on the hydrologic processes and
soil physical properties in an area which
experienced deforestation and conversion
to agricultural land. Cultivated land use
50
Indonesian Journal of Geography, Vol 45, No.1, June 2013 : 48 - 61
create GIS layers [Weng, 2010]. Thematic
information provides descriptive data
about earth surface features and can be
diversified as areas of interest, such as
soil, vegetation, water depth, and land
cover/land use. Thematic information can
be generated through visual interpretation
of remotely sensed data or computer-based
digital image analysis.
has less infiltration rate than forest, thus
causes higher surface runoff and soil loss
rates.
Research based on model simulation on
the hydrological parameters [Pei-Jun Shi
et al., 2007] points out that urbanization
could lead to higher runoff, greater flood
peak discharge and shorter runoff
confluence times, and thus greater risk of
flood
disasters.
The
urbanization
phenomena will increase the impervious
area, which is identified as main factor in
increasing direct runoff [Nie et al., 2011].
According to Davie, [2008] urban areas
have a greater extent of impervious
surfaces than in most natural landforms.
Consequently the amount of infiltration
excess (Hortonian) overland flow is high.
Urban areas are often designed to have a
rapid drainage system, taking the overland
flow away from its source. Where
extensive urbanization of a catchment
occurs, flood frequency and magnitude
increases.
Soil & Water Assessment Tool-Water
Balance (SWAT-WB) is a modified version
of the USDA's Soil & Water Assessment
Tool watershed model (SWAT). This
model uses a physically based soil water
balance to model surface runoff instead of
using the traditional Curve Number
method [White et al., 2009]. The
hydrologic cycle is simulated by the water
balance equation:
and
are the final and
Where,
initial soil water content respectively
(mm), R = daily rainfall (mm), Q = daily
surface runoff (mm), ET = daily
evapotranspiration (mm), P = daily
percolation (mm) and QR = daily lateral
flow (mm).
Remote sensing refers to the activities of
recording, observing, and perceiving
objects or events which in far-away places.
In a more specific definition, remote
sensing is a science and technology to
acquire information about the earth’s
surface and atmosphere using sensors
onboard airborne or space borne platforms
[Weng, 2010].
Arnold et al., [2007] point out that the
original version of SWAT offers two
approaches to determine stream flow;
Curve Number, and the Green-Ampt
routine. To replace the curve number
(CN), SWAT-WB calculates a simple soil
profile water balance for each day of
simulation [Ashagre, 2009]. The soil
moisture routines are used by SWAT-WB
to determine the degree of saturation
deficit for each soil profile for each day of
simulation. This saturation deficit is
termed for the available soil storage
Geographic Information System (GIS) was
a tool of automated mapping and data
management. It has evolved into a capable
spatial
data-handling and
analysis
technology and, more recently, into
geographic information science [Weng,
2010]. The basic concept of GIS is
location and spatial distribution and
relationship of geographic phenomena.
Where is available soil storage, EDC is
the effective depth of the soil profile (unitless), ε is the total soil porosity (mm), and
θ is the volumetric soil moisture for each
day (mm).
The integration of remote sensing and GIS
technologies has been applied widely and
recognized as an effective tool for analysis
[Weng, 2010]. Remotely sensed data can
be used to extract thematic information to
51
ASSESSING THE EFFECTS
Aris Prasena, D.B. Pikha Shrestha
The porosity is a constant value for each
soil type or land mapping unit (HRUs) in
SWAT-WB. Ashagre, [2009] suggests that
by dividing areas into hydrologic response
unit, θ will varies by the day for each of
HRUs and determined by SWAT’s soil
moisture routines. The effective depth,
EDC, a calibration parameter ranging from
zero to one, is used to represent the portion
of the soil profile used in calculating the
saturation deficit.
The available storage, τ, is calculated each
day prior to the start of any rain event.
Once precipitation starts, a portion of the
rain, equal in volume to τ, will infiltrate
the soil. If the rain event is larger in
volume than τ, the soil profile will be
saturated and surface runoff will occur. If
the rain event is less than τ, the soil will
not be saturated and there will be no
surface runoff [White et al., 2009].
Figure 2. Map of Study Area and
Agglomeration of Yogyakarta
The study area has elongated shape since
the value of circularity ratio is 0.2308 [BP
DAS SOP, 2008]. Based on the slope
classes,
Bedog
sub
watershed
predominantly by flat area. Drainage
pattern can be identified by numbers of
river branches. Bedog sub watershed can
be classified as a strong drainage shape
since it has 5-7 branches [BP DAS SOP,
2008].
The study area is Bedog sub watershed
(15528 ha) which is part of Progo
watershed (69648 ha). Bedog is located in
the eastern part of Progo and
administratively lay across two districts in
Yogyakarta Province (Sleman and Bantul)
and the city.
In period of 1986-2006, the largest annual
rainfall in Bedog was 2505 mm/yr
(Ledoknongko Station) and the smallest
annual rainfall was 1740 mm/yr measured
at Patukan Station. Number of wet months
in the study area is 6-8 [Umam, 2010].
The main reason for selection of the study
area is that more than half area of Bedog is
also area of Yogyakarta agglomeration
(figure 2). Spatial planning regulation of
Yogyakarta Province has determined
Yogyakarta agglomeration as a strategic
area and role as national growth pole
[BAPPEDA of Yogyakarta Province,
2010].
Based on the classification of climate
types according to Schmidt and Ferguson
which uses a Q value (ratio between the
average dry months and the average wet
months), the climate types of an area can
be derived. Since Bedog has Q value
between 0.333-0.600 and 0.60-1.00, it can
be concluded that the study area has 2
types of climates namely C (slightly wet)
and D (medium) [BP DAS SOP, 2008].
Umam, [2010] points out that based on
Semi-Detailed Soil Map of Yogyakarta
Special Region (1: 50,000 scale), the study
area consists of 15 types of soil that
classified into sub-groups (figure 3).
52
Indonesian Journal of Geography, Vol 45, No.1, June 2013 : 48 - 61
Based on data from BP DAS SOP,[2008],
complex Red-Brown Latosol - Lithosol
(6183 ha) and Grey-brown Regosol (4845
ha) are predominant soil types in Bedog
sub watershed. Other soil types are;
complex Regosol and Lithosol, Greyishbrown Regosol, complex Lithosol,
Mediterranean and Rendzina, complex
Grey Regosol and Lithosol, and Greybrown Alluvial.
THE METHODS
Land Use
Series of remotely sensed data (2001,
2006, and 2010) were extracted to
generate land use of study area. ASTER
(Advanced Spaceborne Thermal Emission
and Reflection Radiometer) images were
used as data source. The ASTER scenes
covering the study area were projected to
Universal Transverse Mercator (UTM)
coordinate system, Datum WGS 1984,
zone 49 Southern. The images were
corrected to geometric using ground
control points based on topographic map
of Indonesia.
Based on the land use within the study
area we get strongly indication that Bedog
is urbanized sub watershed; there is no
forest, paddy field is predominant land
use, and existence of settlement which is
quite prominent. Alluvial plains and
Fluvio Vulkan plains widely used for
paddy field with irrigation system. Umam,
[2010] argues that mixed garden is another
land use which is quite prominent in
Bedog. Settlements lot located in the
middle of Bedog while mixed garden is a
type of land use that many found in the
upstream and downstream areas that have
a high slope.
Statistical pattern recognition technique
[Jensen, 2005] was an information
extraction method used in this research to
classify land cover/land use. During image
classification on the ASTER images, ENVI
4.5, a software for image processing, was
employed. The previous cloud-free image
was used to extract information when
clouds were found on the image.
The initial step of image classification was
to classify images into classes of land
cover. The supervised classification
technique was employed during training
stage while the maximum likelihood
algorithm was a parametric method used in
classification decision rule [Mather and
Tso, 2009].
Land use map was obtained by intersecting
the result of land cover classification with
the
landform
through
Geographic
Information System (GIS). Results of land
cover classification were exported as
layers in GIS and simple reclassification
method was used to determine land use.
Classification on land use was based on
classification in topographic map of
Indonesia. Land use of study area were
classified into 6 classes namely bare
land/grass, mixed garden, paddy field,
moor/field, settlement, and water body.
Figure 3. Soil Map of Bedog
Source: Umam, 2010
53
ASSESSING THE EFFECTS
Aris Prasena, D.B. Pikha Shrestha
and soil datasets, reclassify layers of land
use and soil, and overlay layers. Once
overlay process was completed, a detail
description of land use and soil
distribution in the sub watershed provided.
Having determined HRUs, SWAT requires
weather data input to simulate runoff.
Change detection analysis was employed
to determine changes in land use.
Conversion, which means a change from
one land use type to another, was the type
of changes that analyzed in this research.
Bi-temporal change detection (direct
comparison) was the approach used in
change detection. It is measure changes
based on a simple timescale comparison
(two data).
Weather Data Inputs
Daily precipitation data were obtained
from the Water Resources Agency (BP
SDA) of Yogyakarta. Based on the rough
screening on the available rainfall data,
there were complete data in period of
2001-2010 that can be used. Consistency
of rainfall data in this research was
analyzed by doing a rough screening on
the data then followed by applying
Spearman’s rank-correlation, and F-test.
Double mass method was employed to
analyze
relative
consistency
and
homogeneity of rainfall data.
Using Arc GIS tools, tables of land use
composition and its changes were resulted.
Pattern of land use for 2001, 2006, and
2010 were analyzed either based on spatial
or percentages of area. Direct comparison
of conversion and its magnitude in period
of 2001-2006 and 2006-2010 were used
for analysis.
SWAT-WB Model Setup Soil and Water
Assessment Tool – Water Balance (SWATWB), a hydrological model, was employed
for this study to simulate runoff based on
information provided by the user. Except
in the runoff calculation, all of processes
in SWAT-WB can be performed through
the interface in Geographic Information
System (GIS) for original SWAT.
ArcSWAT 2009 was the interface that
used in ArcGIS 9.3.
Runoff
The Soil and Water Assessment ToolsWater Balance (SWAT-WB) follows a
saturation excess approach and uses a
simplified water balance. Runoff is
calculated as
Qsurf = P − τ
Where Qsurf is the surface runoff in
millimeters (mm), P is the daily
precipitation in mm, and τ is the available
soil storage in mm.
The main inputs of SWAT model include
Digital Elevation Model (DEM), land use
datasets, soil datasets, and weather data.
The first step required was to delineate
watershed/sub watershed. Automatically
watershed delineation was employed using
the DEM data which obtained from
Planning and Development Agency
(BAPPEDA) of Yogyakarta. Contour map
with 6.25 m contour interval was used as
DEM source.
Avoiding using CN, a simple soil water
balance is calculated by the model for each
day of simulation. SWAT-WB calculates
daily runoff from the amount of rainfall
minus the amount of water that can be
stored in the soil before it is saturated
which called as available soil storage
[White et al, 2009]:
As watershed delineation completed,
determining the hydrologic response units
(HRUs)
which
represents
specific
character of variables that affect runoff
was the further important step in the
research. HRUs were constructed by
define the datasets which were land use
Where is the available soil storage, is
the effective depth of soil profile,
is
the total soil porosity as expressed as a
function of the total soil volume, and
54
Indonesian Journal of Geography, Vol 45, No.1, June 2013 : 48 - 61
Dynamics of Runoff Due to Land Use
Changes
Scenario of land use for the year 2019 was
used to identify the dynamics of runoff
due to land use changes. Land use of 2019
was constructed based on the planning
documents from provincial development
and planning agency (BAPPEDA) of
Yogyakarta. In spatial planning document
of Yogyakarta Province 2009-2028, the
development in Yogyakarta agglomeration
is directed to the settlements, the center of
economic activity and services with
restrict limit to the conversion of fertile
agricultural land. The minimum number of
fertile agricultural land in Yogyakarta
agglomeration should be 20% of the area
in 2019.
is the volumetric soil moisture of
HRU.
Sensitivity Analysis and Calibration
After run the model, the results from the
simulation were evaluated through
sensitivity analysis and calibration in order
to sufficiently predict the runoff. The
purpose of the sensitivity analysis is to
estimate the rate of change in the output of
a model with respect to changes in
[Moriasi et al., 2007]. Sensitivity analysis
was conducted in identifying parameters
that most influential in governing runoff.
The method in the ArcSWAT interface for
sensitivity analysis combines the Latin
Hypercube (LH) sampling and One-factorAt-a-Time (OAT) design for simulation
[van Liew and Veith, 2009]. The
calibration was done by comparing the
average annual conditions from predicted
runoff with the measured data which
derived from discharge. SWAT-WB offers
3 methods of calibration [White et al.,
2009]; Dynamically Dimensioned Search
(DDS) algorithm that used outside of GIS
interface, PARASOL which is an auto
calibration algorithm included in the
SWAT program and available for use
within the GIS interface, and manual
calibration. Manual calibration that based
on the parameters resulted from sensitivity
analysis was employed in this study.
Two runoff prediction scenarios for the
years of 2001 and 2010 using land use of
the year 2019 were conducted to identify
the dynamics of runoff due to land use
changes with respect to the same amount
of rainfall in the study area.
RESULT & DISCUSSION
Land Use Pattern
Based on the coverage of area for each
land use class (table 1), 46.35% of study
area were covered by mixed garden,
24.49%, and 9.51% by paddy field and
settlements respectively. Mixed garden
were found in most parts of sub watershed,
particularly in the south western part and
in the upper part of sub watershed. These
areas have higher slope characteristics and
terrain hilly. Settlements occupied mostly
the middle part of Bedog which
characterized by flat area. Coverage of
paddy field was quite prominent because
of support from irrigation program through
dam establishment and water provision.
The coefficient of determination (R2)
which is the square of the Pearson’s
product-moment correlation coefficient
and the Nash-Sutcliffe Coefficient (NSE)
were used as statistical approaches to
check the model performance [Moriasi et
al., 2007].
Table 1. Land Use of Bedog 2001, 2006, and 2010
Land Use
Bare land/grass
2001
2006
2010
Area (ha)
Area (%)
Area (ha)
Area (%)
Area (ha)
Area (%)
1848.46
13.36
1511.26
10.92
1473.25
10.64
55
ASSESSING THE EFFECTS
Mixed Garden
Aris Prasena, D.B. Pikha Shrestha
6415.09
46.34
6656.20
48.09
5881.28
42.49
Moor/field
848.62
6.13
1104.48
7.98
1046.56
7.56
Paddy Field
3388.98
24.49
2924.68
21.13
3509.68
25.37
Settlement
1316.09
9.51
1622.11
11.72
1909.26
13.79
23.31
0.17
21.82
0.16
20.52
0.15
13840.55
100.00
13840.55
100.00
13840.55
100.00
Water Body
Total
Source: Image Classification, 2011
The patterns of land use distribution in
2006 and 2010 were similar with the
pattern in 2001. Mixed garden occupied
major part of the total land by but there
was an increase of settlements coverage
due to growth of housing demand in the
urban fringe of Yogyakarta. The high
percentage of coverage for mixed garden
and paddy field confirms that agriculture is
the basis of life in Bedog despite the
occurrence of agglomeration process.
Water body was the class with lowest
coverage among the other whilst bare
land/grass covered approximately 10% of
the total area.
Land Use Change
During the period of 2001-2006, the area
of paddy field decreased by 3.35% while
there was expansion of settlements area by
more than 2% (table 2). Area of mixed
garden and moor/field also increased, on
the other side bare land/grass decreased by
more than 2%. The result of change
detection in period of 2001-2006 showed
that 1.55% of the area which was
originally paddy field in 2001 turned into
settlements in 2006.
Table 2. Trend and Magnitude of Land Use Change in Bedog: 2001-2006, and 2006-2010
Land Use
2001-2006
2006-2010
Area (ha)
Area (%)
Area (ha)
Area (%)
Bare land/grass
- 337.20
- 2.44
- 38.01
- 0.27
Mixed Garden
241.11
1.74
- 774.92
- 5.60
Moor/field
255.86
1.85
- 57.91
- 0.42
Paddy Field
- 464.30
- 3.35
584.99
4.23
Settlement
306.02
2.21
287.15
2.07
Water Body
- 1.48
- 0.01
- 1.30
- 0.01
Source: Image Classification, 2011
The period between 2006 and 2010
witnessed a high drop of mixed garden
coverage, on the other side area of paddy
field increased significantly. Settlement
was the only class of land use that
consistently increased for period 20012010. As centre of economic and
education
activities,
Yogyakarta
agglomeration has attracted many people
to move into this area and thus
contributing to the high demand on
settlements as an evidence of physical
expansion of the city into urban fringe
areas.
Based on the change detection result for
period 2006-2010 paddy field coverage
increased despite some areas of paddy
field continually changed into settlements.
This was one of evidences of the success
of program in agriculture and irrigation.
Expansion rate of settlements coverage in
Bedog (0.48% per year) during period of
56
Indonesian Journal of Geography, Vol 45, No.1, June 2013 : 48 - 61
2001 - 2010 is relatively low when
compared to what happened in Yogyakarta
city in general. A research conducted by
Konig et al., [2010] showed that expansion
of building and housing in urban area of
Yogyakarta was 0.81% per year during the
period of 1990-2006. It can be explained
because establishment of new growth
poles in Bedog were less when compared
with other urban fringe in Yogyakarta.
Main growth poles which dominated by
economic and educational centre are
located in the north and east of the city. It
role as trigger in the rapid growth of built
up area.
less sensitive
production.
parameters
in
runoff
Runoff
The model over predicted the runoff for all
of the years. Table 3 showed comparison
of measured and simulated runoff on an
annual basis.
Result of sensitivity analysis was followed
in model calibration which done on annual
basis. Runoff calibration for the study area
was conducted for the years 2001, 2006,
and 2010. Data of previous years
respectively were used for warm up
simulation. Calibration was performed
manually until the predicted values meet
with the observed annual average.
Available water capacity of the soil layer
(SOL_AWC) increased by 12.5% in order
to reduce surface runoff, soil evaporation
compensation factor (ESCO) was adjusted
into 0.8 in attempt to decrease total flow.
The effective depth coefficient (EDC)
value was adjusted to be 0.985. The result
on average of annual surface run off after
calibration was shown in the table 4.
Table 3. Comparison of Observed and
Predicted Runoff
Table 4. Average of Annual Observed and
Predicted Runoff after Calibration
Runoff
Observed
(mm)
Predicted
(mm)
2001
2006
2010
1424.28
988.22
1457.26
1749.77
1216.50
1870.82
Runoff
Observed
(mm)
Predicted
(mm)
Soil properties; available water capacity
soil
evaporation
(SOL_AWC),
compensation factor (ESCO), and depth
from soil surface to bottom of layer
(SOL_Z) in mm) were the most sensitive
parameter in runoff production. The
available water capacity (SOL_AWC) in
mm water / mm soil was found to be the
most sensitive parameter in sensitivity
analysis.
2001
2006
2010
1424.28
988.22
1457.26
1559.86
1057.43
1525.00
The predicted monthly runoff was lower
than the observed at the first rainy season
(February-May) in 2001. On the other
side, the predicted run off was higher than
the observed data at the second the rainy
season (October-December). In 2006 and
2010, the model predicted runoff higher
than observed data in the first wet season
while under predicted in the second wet
season. Difference on the pattern of the
model to predict runoff at the beginning of
rainy season can be occurred by the
influence of the available initial soil
moisture from previous year.
Runoff generation was also found to be
sensitive to groundwater parameters;
threshold depth of water in the shallow
aquifer required for return flow to occur
(GWQMN) in mm, and the base flow
alpha factor (ALPHA_BF) in days. The
crop parameters; maximum canopy storage
(CANMX) in mm H2O, maximum
potential leaf area index (BLAI), and plant
uptake compensation factor (EPCO) were
The performance of SWAT-WB was
objectively evaluated by coefficient of
determination (R2) and Nash-Sutcliffe
coefficient of Efficiency (NSE). The
results of statistical parameter used for
57
ASSESSING THE EFFECTS
model performance
tabulated in table 5.
Aris Prasena, D.B. Pikha Shrestha
evaluation
settlements in the study area increased
from 9.51% in 2001 to 13.79%, 22.72% in
2010, 2019 respectively.
were
Table 5. Result of Model Evaluation
Performance
Statistical
Parameter
R2
NSE
2001
0.70
0.64
Year
2006
0.57
0.5 [Moriasi et al., 2007]. The
statistical results for NSE revealed that the
mean observed value in 2006 was a better
predictor than the predicted value which
indicates unacceptable performance. On
the other hand, results for 2001 and 2010
were generally viewed as acceptable
performance [Moriasi et al., 2007].
Increase in runoff for 4.67% and 3.42% as
the contribution from the expansion of
13.21% and 8.93% in settlements coverage
can be classified as small quantities. A
study using SWAT in Malaysia [Alansi et
al., 2009] revealed the expansion of the
settlement for 13.17% led to changes in
runoff reaches 8%.
Table 6. Land Use Scenario 2019
Effect of Land Use Change on Runoff
A land use scenario was established in
order to investigate the dynamics of runoff
due to changes in land use. The model was
re-run using the land use scenario and
weather input used in 2001 and 2010. The
scenario of land use in 2019 was
constructed based on spatial planning
document of Yogyakarta 2009-2028. The
composition of land use scenario was
shown by (table 6). The coverage of
Land Use
2019
Area (ha)
Area (%)
855.26
6.18
5654.47
40.85
Moor/field
306.05
2.21
Paddy Field
3859.67
27.89
Settlement
3144.58
22.72
20.52
0.15
13840.55
100.00
Bare land/grass
Mixed garden
Water Body
Total
Table 7. Comparison of Runoff Land Use 2001-2010 on Runoff Land Use Scenario
Month
January
February
March
April
May
June
July
August
September
October
November
December
LU 2001
162.37
150.98
323.42
71.60
46.88
118.82
7.96
0
2.59
209.23
407.50
59.05
1560.40
Runoff (mm)
LU Scenario
LU 2010
172.46
140.28
160.54
148.20
339.30
177.88
75.69
54.77
50.34
195.57
122.79
111.33
7.93
28.04
0
60.73
2.51
28.20
219.60
126.94
420.11
146.89
62.04
306.06
1633.31
1524.89
58
LU Scenario
145.58
152.77
183.23
55.45
201.26
113.15
27.69
65.10
37.80
129.75
151.61
313.60
1576.99
Indonesian Journal of Geography, Vol 45, No.1, June 2013 : 48 - 61
respectively, but it was unacceptable
for 2006 with the value
IJG Vol. 45, No.1, June 2013 (48 - 61)
© 2013 Faculty of Geography UGM and
The Indonesian Geographers Association
ASSESSING THE EFFECTS OF LAND USE CHANGE ON RUNOFF
IN BEDOG SUB WATERSHED YOGYAKARTA
Aris Prasena
prasena.aris@gmail.com
BAPPRDA PROV DIY kompleks Kepatihan Danurejan, Yogyakarta
D.B. Pikha Shrestha
shrestha@itc.nl
Faculty of Geo-Information and Earth Observation, University Of Twente
ABSTRACT
The study was to assess the effects of land use change on runoff in the Bedog sub watershed.
Soil and Water Assessment Tool-Water Balance (SWAT-WB) hydrological modeling was used
to predict runoff for years of 2001, 2006, and 2010. Land use in Bedog has rapidly changed
in last few decades due to agglomeration process in Yogyakarta City. Coverage of mixed
garden decreased during period of 2001-2010 despite still as predominant land use in total.
On the other side, change detection analysis revealed that there was an increase of
settlements coverage from 9.51% to 13.79% in the same period. Sensitivity analysis revealed
that soil properties were the most sensitive parameters on runoff generation. Calibration was
performed for years of 2001, 2006, and 2010 and the result shows an acceptable
performance in runoff simulation. Changes in land use were responsible for an increase in
the annual runoff between 3.42% 4.67%. This study showed that dynamics of runoff can be
predicted by forecasting and simulating future land use.
Key Words: Remote sensing, Geographic Information System (GIS), land use change, Soil and
Water Assessment Tool-Water Balance (SWAT-WB), runoff.
ABSTRAK
Penelitian ini bertujuan untuk menganalisis dampak dari perubahan penggunaan lahan
terhadap limpasan di sub DAS Bedog. Pemodelan hidrologi berupa Soil and Water
Assessment Tool-Water Balance (SWAT-WB) digunakan untuk memprediksi limpasan
selama tahun 2001, 2006, dan 2010. Penggunaan lahan di Bedog dengan cepat berubah
dalam beberapa dekade terakhir akibat proses aglomerasi di Kota Yogyakarta. Lahan
berupa kebun campuran berkurang selama periode 2001-2010, meskipun lahan tersebut
masih merupakan penggunaan lahan dominan secara keseluruhan. Di sisi lain, analisis
deteksi perubahan menunjukan bahwa terjadi peningkatan cakupan permukiman dari 9,51%
menjadi 13,79% pada periode yang sama. Analisis sensitivitas menunjukkan bahwa sifat
tanah merupakan parameter yang paling sensitif dalam menghasilkan limpasan. Kalibrasi
yang dilakukan selama bertahun-tahun 2001, 2006, dan 2010 menunjukkan kinerja yang
dapat diterima dalam simulasi limpasan. Perubahan penggunaan lahan berdampak pada
peningkatan limpasan tahunan antara 3,42% - 4,67%. Studi ini menunjukkan bahwa
dinamika limpasan dapat diprediksi dengan peramalan dan simulasi penggunaan lahan di
masa depan.
Kata Kunci: Penginderaan jauh, Sistem Informasi Geografis (GIS), perubahan penggunaan
lahan, Soil and Water Assessment Tool-Water Balance (SWAT-WB), limpasan.
Indonesian Journal of Geography, Vol 45, No.1, June 2013 : 48 - 61
INTRODUCTION
Changes in land use and land cover have
occurred from prehistoric times till present
[Ellis, 2010]. Mankind has changed land
surface in attempts to improve the
availability of space, and enhance security
of basic natural resources. The phenomena
of land use changes and its consequences
to the environment are important to be
observed in attempts to implementing
sustainable development.
Agglomeration of Yogyakarta as the
centre of economic activities and
education services in Yogyakarta Province
has witnessed rapid development of
buildings, infrastructure, and changes of
land use from non-urban to urban features.
As a consequence land use change took
place, growth of settlement in Bantul,
Sleman, and Yogyakarta increased 7.16%
in period of 1994-2000 while on the other
hand areas of paddy field decreased from
46.31 % to 39.49 % [BAPPEDA of
Yogyakarta Province, 2007].
Figure 1. Map of Progo watershed
As consequence, extensive changes from
non-urban into urban features possibly
generate higher runoff that will increase
flood risk both in the area of Yogyakarta
agglomeration and in the lower part of
watershed.
Bedog is part of Progo watershed which
one of urbanized watersheds (figure 1).
The middle stream of Bedog is part of
agglomeration
Yogyakarta
which
determined as national growth pole. It has
role as centre of economic and education
activities. On the other hand, most part of
upper area and downstream has different
characteristic as an agriculture areas that
contribute in food security program.
The main objective of this research is to
assess the effects of land use change on
runoff in Bedog sub watershed,
Yogyakarta Province, Indonesia.
The specific objectives of this research
are:
1. To identify the existing land use
pattern.
2. To analyze pattern and magnitude
of land use change 2001-2010.
3. To determine runoff using a model.
4. To analyze the relation between
land use change and runoff
dynamic.
Middle part of Bedog as urban fringe of
Yogyakarta has witnessed urbanization
and changes in land use.
Entering the 20th century, land use change
has become the world’s problem closely
related with other global issues such as
population growth, food security, and
climate change. It is estimated that builtup or impervious areas occupy between
2% to 3% of the Earth’s land surface
49
ASSESSING THE EFFECTS
Aris Prasena, D.B. Pikha Shrestha
vegetation, soil, lakes, rivers, snowfields
and oceans.
[Lambin and Geist, 2006]. This is also
supported by deforestation in tropical
countries [Pagiola, 2000] and extensively
application of agricultural intensification
in productive land using fertilizers,
pesticides and irrigation systems in recent
decades.
Precipitation which covers all forms of
water being released by the atmosphere is
the beginning of a whole chain of events
that occurs in watersheds and the source of
water that replenishes soil moisture,
stream flows, lakes, glaciers, etc [Salas,
2006]. According to Washington State
Department
of
Ecology,
[1986]
precipitation falls on the earth and either
percolates into the soil or flows across the
ground.
In Indonesia, the loss of natural forest in
Sumatra, Kalimantan, Sulawesi, and Irian
Jaya between 1985 and 1997 is over 19
million ha of forest, including 6.7 million
ha in Sumatra and 8.5 million ha in
Kalimantan [Pagiola, 2000]. This amount
is an average annual rate in those two
islands of 1.26 million ha per year.
The term runoff incorporates the
movement of liquid water above and
below the surface of the earth [Davie,
2008]. Surface runoff is part of the rainfall
that flows over the surface of the soil to
the rivers, lakes, and oceans [Asdak,
1995]. Runoff mechanisms that contribute
to stream flow can be distinguished as
overland flow, through flow/lateral flow,
and groundwater flow [Davie, 2008].
Overland flow is the water which runs
across the surface of the land before
reaching the stream. Lateral flow is the
water which runs the subsurface, occurs in
the shallow subsurface, predominantly,
although not always, in the unsaturated
zone. Groundwater flow is in the deeper
saturated zone.
Yunus,
[2008]
points
out
that
industrialization has become main driving
force for rapid urbanization since most of
manufacturing industries tend to be
located around city centre to minimize
production cost. Lambin and Geist, [2006]
argue that urbanization affects land in rural
areas through the ecological footprint of
cities includes, agricultural land in periurban areas for residential, infrastructure,
and amenity uses, which blurs the
distinction between cities and rural.
Land use change due to urban growth has
occurred in Yogyakarta Province. Konig et
al., [2010] point out that population
density in the Yogyakarta increased by
84% (from 532 to 979 persons per km2) in
1970-2000. This rapid growth resulted in
urban-rural expansions of built-up areas by
13% (1990–2006). In period of 19932006, growth of built-up areas and new
rural settlements was doubled, while
agricultural land decreased by 25%.
Factors affecting runoff are factors related
to climate, particularly rainfall and
characteristics of watershed [Asdak, 1995].
The rate and volume of runoff is
determined by duration, intensity, and
distribution of rainfall while characteristics
of watershed are shape, morphometry,
topography, geology, and land use (type
and vegetation density).
Hydrological cycle is a conceptual model
of how water moves around between the
earth and atmosphere in different states as
a gas, liquid or solid [Davie, 2008].
Washington State Department of Ecology
[1986] suggests that the hydrologic cycle
begins when water evaporates into the
atmosphere that can be derived from
Land use change plays role as main factor
in hydrological cycle. Giertz et al., [2004]
point out that there are effects of land use
change on the hydrologic processes and
soil physical properties in an area which
experienced deforestation and conversion
to agricultural land. Cultivated land use
50
Indonesian Journal of Geography, Vol 45, No.1, June 2013 : 48 - 61
create GIS layers [Weng, 2010]. Thematic
information provides descriptive data
about earth surface features and can be
diversified as areas of interest, such as
soil, vegetation, water depth, and land
cover/land use. Thematic information can
be generated through visual interpretation
of remotely sensed data or computer-based
digital image analysis.
has less infiltration rate than forest, thus
causes higher surface runoff and soil loss
rates.
Research based on model simulation on
the hydrological parameters [Pei-Jun Shi
et al., 2007] points out that urbanization
could lead to higher runoff, greater flood
peak discharge and shorter runoff
confluence times, and thus greater risk of
flood
disasters.
The
urbanization
phenomena will increase the impervious
area, which is identified as main factor in
increasing direct runoff [Nie et al., 2011].
According to Davie, [2008] urban areas
have a greater extent of impervious
surfaces than in most natural landforms.
Consequently the amount of infiltration
excess (Hortonian) overland flow is high.
Urban areas are often designed to have a
rapid drainage system, taking the overland
flow away from its source. Where
extensive urbanization of a catchment
occurs, flood frequency and magnitude
increases.
Soil & Water Assessment Tool-Water
Balance (SWAT-WB) is a modified version
of the USDA's Soil & Water Assessment
Tool watershed model (SWAT). This
model uses a physically based soil water
balance to model surface runoff instead of
using the traditional Curve Number
method [White et al., 2009]. The
hydrologic cycle is simulated by the water
balance equation:
and
are the final and
Where,
initial soil water content respectively
(mm), R = daily rainfall (mm), Q = daily
surface runoff (mm), ET = daily
evapotranspiration (mm), P = daily
percolation (mm) and QR = daily lateral
flow (mm).
Remote sensing refers to the activities of
recording, observing, and perceiving
objects or events which in far-away places.
In a more specific definition, remote
sensing is a science and technology to
acquire information about the earth’s
surface and atmosphere using sensors
onboard airborne or space borne platforms
[Weng, 2010].
Arnold et al., [2007] point out that the
original version of SWAT offers two
approaches to determine stream flow;
Curve Number, and the Green-Ampt
routine. To replace the curve number
(CN), SWAT-WB calculates a simple soil
profile water balance for each day of
simulation [Ashagre, 2009]. The soil
moisture routines are used by SWAT-WB
to determine the degree of saturation
deficit for each soil profile for each day of
simulation. This saturation deficit is
termed for the available soil storage
Geographic Information System (GIS) was
a tool of automated mapping and data
management. It has evolved into a capable
spatial
data-handling and
analysis
technology and, more recently, into
geographic information science [Weng,
2010]. The basic concept of GIS is
location and spatial distribution and
relationship of geographic phenomena.
Where is available soil storage, EDC is
the effective depth of the soil profile (unitless), ε is the total soil porosity (mm), and
θ is the volumetric soil moisture for each
day (mm).
The integration of remote sensing and GIS
technologies has been applied widely and
recognized as an effective tool for analysis
[Weng, 2010]. Remotely sensed data can
be used to extract thematic information to
51
ASSESSING THE EFFECTS
Aris Prasena, D.B. Pikha Shrestha
The porosity is a constant value for each
soil type or land mapping unit (HRUs) in
SWAT-WB. Ashagre, [2009] suggests that
by dividing areas into hydrologic response
unit, θ will varies by the day for each of
HRUs and determined by SWAT’s soil
moisture routines. The effective depth,
EDC, a calibration parameter ranging from
zero to one, is used to represent the portion
of the soil profile used in calculating the
saturation deficit.
The available storage, τ, is calculated each
day prior to the start of any rain event.
Once precipitation starts, a portion of the
rain, equal in volume to τ, will infiltrate
the soil. If the rain event is larger in
volume than τ, the soil profile will be
saturated and surface runoff will occur. If
the rain event is less than τ, the soil will
not be saturated and there will be no
surface runoff [White et al., 2009].
Figure 2. Map of Study Area and
Agglomeration of Yogyakarta
The study area has elongated shape since
the value of circularity ratio is 0.2308 [BP
DAS SOP, 2008]. Based on the slope
classes,
Bedog
sub
watershed
predominantly by flat area. Drainage
pattern can be identified by numbers of
river branches. Bedog sub watershed can
be classified as a strong drainage shape
since it has 5-7 branches [BP DAS SOP,
2008].
The study area is Bedog sub watershed
(15528 ha) which is part of Progo
watershed (69648 ha). Bedog is located in
the eastern part of Progo and
administratively lay across two districts in
Yogyakarta Province (Sleman and Bantul)
and the city.
In period of 1986-2006, the largest annual
rainfall in Bedog was 2505 mm/yr
(Ledoknongko Station) and the smallest
annual rainfall was 1740 mm/yr measured
at Patukan Station. Number of wet months
in the study area is 6-8 [Umam, 2010].
The main reason for selection of the study
area is that more than half area of Bedog is
also area of Yogyakarta agglomeration
(figure 2). Spatial planning regulation of
Yogyakarta Province has determined
Yogyakarta agglomeration as a strategic
area and role as national growth pole
[BAPPEDA of Yogyakarta Province,
2010].
Based on the classification of climate
types according to Schmidt and Ferguson
which uses a Q value (ratio between the
average dry months and the average wet
months), the climate types of an area can
be derived. Since Bedog has Q value
between 0.333-0.600 and 0.60-1.00, it can
be concluded that the study area has 2
types of climates namely C (slightly wet)
and D (medium) [BP DAS SOP, 2008].
Umam, [2010] points out that based on
Semi-Detailed Soil Map of Yogyakarta
Special Region (1: 50,000 scale), the study
area consists of 15 types of soil that
classified into sub-groups (figure 3).
52
Indonesian Journal of Geography, Vol 45, No.1, June 2013 : 48 - 61
Based on data from BP DAS SOP,[2008],
complex Red-Brown Latosol - Lithosol
(6183 ha) and Grey-brown Regosol (4845
ha) are predominant soil types in Bedog
sub watershed. Other soil types are;
complex Regosol and Lithosol, Greyishbrown Regosol, complex Lithosol,
Mediterranean and Rendzina, complex
Grey Regosol and Lithosol, and Greybrown Alluvial.
THE METHODS
Land Use
Series of remotely sensed data (2001,
2006, and 2010) were extracted to
generate land use of study area. ASTER
(Advanced Spaceborne Thermal Emission
and Reflection Radiometer) images were
used as data source. The ASTER scenes
covering the study area were projected to
Universal Transverse Mercator (UTM)
coordinate system, Datum WGS 1984,
zone 49 Southern. The images were
corrected to geometric using ground
control points based on topographic map
of Indonesia.
Based on the land use within the study
area we get strongly indication that Bedog
is urbanized sub watershed; there is no
forest, paddy field is predominant land
use, and existence of settlement which is
quite prominent. Alluvial plains and
Fluvio Vulkan plains widely used for
paddy field with irrigation system. Umam,
[2010] argues that mixed garden is another
land use which is quite prominent in
Bedog. Settlements lot located in the
middle of Bedog while mixed garden is a
type of land use that many found in the
upstream and downstream areas that have
a high slope.
Statistical pattern recognition technique
[Jensen, 2005] was an information
extraction method used in this research to
classify land cover/land use. During image
classification on the ASTER images, ENVI
4.5, a software for image processing, was
employed. The previous cloud-free image
was used to extract information when
clouds were found on the image.
The initial step of image classification was
to classify images into classes of land
cover. The supervised classification
technique was employed during training
stage while the maximum likelihood
algorithm was a parametric method used in
classification decision rule [Mather and
Tso, 2009].
Land use map was obtained by intersecting
the result of land cover classification with
the
landform
through
Geographic
Information System (GIS). Results of land
cover classification were exported as
layers in GIS and simple reclassification
method was used to determine land use.
Classification on land use was based on
classification in topographic map of
Indonesia. Land use of study area were
classified into 6 classes namely bare
land/grass, mixed garden, paddy field,
moor/field, settlement, and water body.
Figure 3. Soil Map of Bedog
Source: Umam, 2010
53
ASSESSING THE EFFECTS
Aris Prasena, D.B. Pikha Shrestha
and soil datasets, reclassify layers of land
use and soil, and overlay layers. Once
overlay process was completed, a detail
description of land use and soil
distribution in the sub watershed provided.
Having determined HRUs, SWAT requires
weather data input to simulate runoff.
Change detection analysis was employed
to determine changes in land use.
Conversion, which means a change from
one land use type to another, was the type
of changes that analyzed in this research.
Bi-temporal change detection (direct
comparison) was the approach used in
change detection. It is measure changes
based on a simple timescale comparison
(two data).
Weather Data Inputs
Daily precipitation data were obtained
from the Water Resources Agency (BP
SDA) of Yogyakarta. Based on the rough
screening on the available rainfall data,
there were complete data in period of
2001-2010 that can be used. Consistency
of rainfall data in this research was
analyzed by doing a rough screening on
the data then followed by applying
Spearman’s rank-correlation, and F-test.
Double mass method was employed to
analyze
relative
consistency
and
homogeneity of rainfall data.
Using Arc GIS tools, tables of land use
composition and its changes were resulted.
Pattern of land use for 2001, 2006, and
2010 were analyzed either based on spatial
or percentages of area. Direct comparison
of conversion and its magnitude in period
of 2001-2006 and 2006-2010 were used
for analysis.
SWAT-WB Model Setup Soil and Water
Assessment Tool – Water Balance (SWATWB), a hydrological model, was employed
for this study to simulate runoff based on
information provided by the user. Except
in the runoff calculation, all of processes
in SWAT-WB can be performed through
the interface in Geographic Information
System (GIS) for original SWAT.
ArcSWAT 2009 was the interface that
used in ArcGIS 9.3.
Runoff
The Soil and Water Assessment ToolsWater Balance (SWAT-WB) follows a
saturation excess approach and uses a
simplified water balance. Runoff is
calculated as
Qsurf = P − τ
Where Qsurf is the surface runoff in
millimeters (mm), P is the daily
precipitation in mm, and τ is the available
soil storage in mm.
The main inputs of SWAT model include
Digital Elevation Model (DEM), land use
datasets, soil datasets, and weather data.
The first step required was to delineate
watershed/sub watershed. Automatically
watershed delineation was employed using
the DEM data which obtained from
Planning and Development Agency
(BAPPEDA) of Yogyakarta. Contour map
with 6.25 m contour interval was used as
DEM source.
Avoiding using CN, a simple soil water
balance is calculated by the model for each
day of simulation. SWAT-WB calculates
daily runoff from the amount of rainfall
minus the amount of water that can be
stored in the soil before it is saturated
which called as available soil storage
[White et al, 2009]:
As watershed delineation completed,
determining the hydrologic response units
(HRUs)
which
represents
specific
character of variables that affect runoff
was the further important step in the
research. HRUs were constructed by
define the datasets which were land use
Where is the available soil storage, is
the effective depth of soil profile,
is
the total soil porosity as expressed as a
function of the total soil volume, and
54
Indonesian Journal of Geography, Vol 45, No.1, June 2013 : 48 - 61
Dynamics of Runoff Due to Land Use
Changes
Scenario of land use for the year 2019 was
used to identify the dynamics of runoff
due to land use changes. Land use of 2019
was constructed based on the planning
documents from provincial development
and planning agency (BAPPEDA) of
Yogyakarta. In spatial planning document
of Yogyakarta Province 2009-2028, the
development in Yogyakarta agglomeration
is directed to the settlements, the center of
economic activity and services with
restrict limit to the conversion of fertile
agricultural land. The minimum number of
fertile agricultural land in Yogyakarta
agglomeration should be 20% of the area
in 2019.
is the volumetric soil moisture of
HRU.
Sensitivity Analysis and Calibration
After run the model, the results from the
simulation were evaluated through
sensitivity analysis and calibration in order
to sufficiently predict the runoff. The
purpose of the sensitivity analysis is to
estimate the rate of change in the output of
a model with respect to changes in
[Moriasi et al., 2007]. Sensitivity analysis
was conducted in identifying parameters
that most influential in governing runoff.
The method in the ArcSWAT interface for
sensitivity analysis combines the Latin
Hypercube (LH) sampling and One-factorAt-a-Time (OAT) design for simulation
[van Liew and Veith, 2009]. The
calibration was done by comparing the
average annual conditions from predicted
runoff with the measured data which
derived from discharge. SWAT-WB offers
3 methods of calibration [White et al.,
2009]; Dynamically Dimensioned Search
(DDS) algorithm that used outside of GIS
interface, PARASOL which is an auto
calibration algorithm included in the
SWAT program and available for use
within the GIS interface, and manual
calibration. Manual calibration that based
on the parameters resulted from sensitivity
analysis was employed in this study.
Two runoff prediction scenarios for the
years of 2001 and 2010 using land use of
the year 2019 were conducted to identify
the dynamics of runoff due to land use
changes with respect to the same amount
of rainfall in the study area.
RESULT & DISCUSSION
Land Use Pattern
Based on the coverage of area for each
land use class (table 1), 46.35% of study
area were covered by mixed garden,
24.49%, and 9.51% by paddy field and
settlements respectively. Mixed garden
were found in most parts of sub watershed,
particularly in the south western part and
in the upper part of sub watershed. These
areas have higher slope characteristics and
terrain hilly. Settlements occupied mostly
the middle part of Bedog which
characterized by flat area. Coverage of
paddy field was quite prominent because
of support from irrigation program through
dam establishment and water provision.
The coefficient of determination (R2)
which is the square of the Pearson’s
product-moment correlation coefficient
and the Nash-Sutcliffe Coefficient (NSE)
were used as statistical approaches to
check the model performance [Moriasi et
al., 2007].
Table 1. Land Use of Bedog 2001, 2006, and 2010
Land Use
Bare land/grass
2001
2006
2010
Area (ha)
Area (%)
Area (ha)
Area (%)
Area (ha)
Area (%)
1848.46
13.36
1511.26
10.92
1473.25
10.64
55
ASSESSING THE EFFECTS
Mixed Garden
Aris Prasena, D.B. Pikha Shrestha
6415.09
46.34
6656.20
48.09
5881.28
42.49
Moor/field
848.62
6.13
1104.48
7.98
1046.56
7.56
Paddy Field
3388.98
24.49
2924.68
21.13
3509.68
25.37
Settlement
1316.09
9.51
1622.11
11.72
1909.26
13.79
23.31
0.17
21.82
0.16
20.52
0.15
13840.55
100.00
13840.55
100.00
13840.55
100.00
Water Body
Total
Source: Image Classification, 2011
The patterns of land use distribution in
2006 and 2010 were similar with the
pattern in 2001. Mixed garden occupied
major part of the total land by but there
was an increase of settlements coverage
due to growth of housing demand in the
urban fringe of Yogyakarta. The high
percentage of coverage for mixed garden
and paddy field confirms that agriculture is
the basis of life in Bedog despite the
occurrence of agglomeration process.
Water body was the class with lowest
coverage among the other whilst bare
land/grass covered approximately 10% of
the total area.
Land Use Change
During the period of 2001-2006, the area
of paddy field decreased by 3.35% while
there was expansion of settlements area by
more than 2% (table 2). Area of mixed
garden and moor/field also increased, on
the other side bare land/grass decreased by
more than 2%. The result of change
detection in period of 2001-2006 showed
that 1.55% of the area which was
originally paddy field in 2001 turned into
settlements in 2006.
Table 2. Trend and Magnitude of Land Use Change in Bedog: 2001-2006, and 2006-2010
Land Use
2001-2006
2006-2010
Area (ha)
Area (%)
Area (ha)
Area (%)
Bare land/grass
- 337.20
- 2.44
- 38.01
- 0.27
Mixed Garden
241.11
1.74
- 774.92
- 5.60
Moor/field
255.86
1.85
- 57.91
- 0.42
Paddy Field
- 464.30
- 3.35
584.99
4.23
Settlement
306.02
2.21
287.15
2.07
Water Body
- 1.48
- 0.01
- 1.30
- 0.01
Source: Image Classification, 2011
The period between 2006 and 2010
witnessed a high drop of mixed garden
coverage, on the other side area of paddy
field increased significantly. Settlement
was the only class of land use that
consistently increased for period 20012010. As centre of economic and
education
activities,
Yogyakarta
agglomeration has attracted many people
to move into this area and thus
contributing to the high demand on
settlements as an evidence of physical
expansion of the city into urban fringe
areas.
Based on the change detection result for
period 2006-2010 paddy field coverage
increased despite some areas of paddy
field continually changed into settlements.
This was one of evidences of the success
of program in agriculture and irrigation.
Expansion rate of settlements coverage in
Bedog (0.48% per year) during period of
56
Indonesian Journal of Geography, Vol 45, No.1, June 2013 : 48 - 61
2001 - 2010 is relatively low when
compared to what happened in Yogyakarta
city in general. A research conducted by
Konig et al., [2010] showed that expansion
of building and housing in urban area of
Yogyakarta was 0.81% per year during the
period of 1990-2006. It can be explained
because establishment of new growth
poles in Bedog were less when compared
with other urban fringe in Yogyakarta.
Main growth poles which dominated by
economic and educational centre are
located in the north and east of the city. It
role as trigger in the rapid growth of built
up area.
less sensitive
production.
parameters
in
runoff
Runoff
The model over predicted the runoff for all
of the years. Table 3 showed comparison
of measured and simulated runoff on an
annual basis.
Result of sensitivity analysis was followed
in model calibration which done on annual
basis. Runoff calibration for the study area
was conducted for the years 2001, 2006,
and 2010. Data of previous years
respectively were used for warm up
simulation. Calibration was performed
manually until the predicted values meet
with the observed annual average.
Available water capacity of the soil layer
(SOL_AWC) increased by 12.5% in order
to reduce surface runoff, soil evaporation
compensation factor (ESCO) was adjusted
into 0.8 in attempt to decrease total flow.
The effective depth coefficient (EDC)
value was adjusted to be 0.985. The result
on average of annual surface run off after
calibration was shown in the table 4.
Table 3. Comparison of Observed and
Predicted Runoff
Table 4. Average of Annual Observed and
Predicted Runoff after Calibration
Runoff
Observed
(mm)
Predicted
(mm)
2001
2006
2010
1424.28
988.22
1457.26
1749.77
1216.50
1870.82
Runoff
Observed
(mm)
Predicted
(mm)
Soil properties; available water capacity
soil
evaporation
(SOL_AWC),
compensation factor (ESCO), and depth
from soil surface to bottom of layer
(SOL_Z) in mm) were the most sensitive
parameter in runoff production. The
available water capacity (SOL_AWC) in
mm water / mm soil was found to be the
most sensitive parameter in sensitivity
analysis.
2001
2006
2010
1424.28
988.22
1457.26
1559.86
1057.43
1525.00
The predicted monthly runoff was lower
than the observed at the first rainy season
(February-May) in 2001. On the other
side, the predicted run off was higher than
the observed data at the second the rainy
season (October-December). In 2006 and
2010, the model predicted runoff higher
than observed data in the first wet season
while under predicted in the second wet
season. Difference on the pattern of the
model to predict runoff at the beginning of
rainy season can be occurred by the
influence of the available initial soil
moisture from previous year.
Runoff generation was also found to be
sensitive to groundwater parameters;
threshold depth of water in the shallow
aquifer required for return flow to occur
(GWQMN) in mm, and the base flow
alpha factor (ALPHA_BF) in days. The
crop parameters; maximum canopy storage
(CANMX) in mm H2O, maximum
potential leaf area index (BLAI), and plant
uptake compensation factor (EPCO) were
The performance of SWAT-WB was
objectively evaluated by coefficient of
determination (R2) and Nash-Sutcliffe
coefficient of Efficiency (NSE). The
results of statistical parameter used for
57
ASSESSING THE EFFECTS
model performance
tabulated in table 5.
Aris Prasena, D.B. Pikha Shrestha
evaluation
settlements in the study area increased
from 9.51% in 2001 to 13.79%, 22.72% in
2010, 2019 respectively.
were
Table 5. Result of Model Evaluation
Performance
Statistical
Parameter
R2
NSE
2001
0.70
0.64
Year
2006
0.57
0.5 [Moriasi et al., 2007]. The
statistical results for NSE revealed that the
mean observed value in 2006 was a better
predictor than the predicted value which
indicates unacceptable performance. On
the other hand, results for 2001 and 2010
were generally viewed as acceptable
performance [Moriasi et al., 2007].
Increase in runoff for 4.67% and 3.42% as
the contribution from the expansion of
13.21% and 8.93% in settlements coverage
can be classified as small quantities. A
study using SWAT in Malaysia [Alansi et
al., 2009] revealed the expansion of the
settlement for 13.17% led to changes in
runoff reaches 8%.
Table 6. Land Use Scenario 2019
Effect of Land Use Change on Runoff
A land use scenario was established in
order to investigate the dynamics of runoff
due to changes in land use. The model was
re-run using the land use scenario and
weather input used in 2001 and 2010. The
scenario of land use in 2019 was
constructed based on spatial planning
document of Yogyakarta 2009-2028. The
composition of land use scenario was
shown by (table 6). The coverage of
Land Use
2019
Area (ha)
Area (%)
855.26
6.18
5654.47
40.85
Moor/field
306.05
2.21
Paddy Field
3859.67
27.89
Settlement
3144.58
22.72
20.52
0.15
13840.55
100.00
Bare land/grass
Mixed garden
Water Body
Total
Table 7. Comparison of Runoff Land Use 2001-2010 on Runoff Land Use Scenario
Month
January
February
March
April
May
June
July
August
September
October
November
December
LU 2001
162.37
150.98
323.42
71.60
46.88
118.82
7.96
0
2.59
209.23
407.50
59.05
1560.40
Runoff (mm)
LU Scenario
LU 2010
172.46
140.28
160.54
148.20
339.30
177.88
75.69
54.77
50.34
195.57
122.79
111.33
7.93
28.04
0
60.73
2.51
28.20
219.60
126.94
420.11
146.89
62.04
306.06
1633.31
1524.89
58
LU Scenario
145.58
152.77
183.23
55.45
201.26
113.15
27.69
65.10
37.80
129.75
151.61
313.60
1576.99
Indonesian Journal of Geography, Vol 45, No.1, June 2013 : 48 - 61
respectively, but it was unacceptable
for 2006 with the value